70 likes | 211 Views
SST and sea ice data - recommendations to Historical Marine Data workshop. Sea ice thickness information required for model heat fluxes historical Russian (AARI) data should be investigated and incorporated into Arctic time series if possible SST
E N D
SST and sea ice data - recommendations to Historical Marine Data workshop • Sea ice • thickness information required for model heat fluxes • historical Russian (AARI) data should be investigated and incorporated into Arctic time series if possible • SST • as model resolutions increase, data for worldwide inland seas and large lakes (including historical information) needed including estimates of freezing points and evaporation rates. • archive quality controlled original SST measurements for assimilation into/nudging CGCMs • assemble sets of skin SSTs to test model sensitivity to these (or is this a modelling problem?). Problem may be in tropical diurnal cycle and associated convection • document modes of variability (EOFs) included in analyses, their relative contributions and their goodness of fit through time
SST and sea ice data - recommendations to Historical Marine Data workshop • SST • provide all data sets/analyses with estimates of error associated with each grid box (no methods should be used that are not capable of delivering these). • some modification to HadISST probably needed • tests of the sensitivity of AGCMs to alternative SSTs are required, either by perturbing analyses using their error estimates or using multiple analyses. Second stage of C20C. • can we produce sub-monthly historical analyses?
In addition, how should HadISST be improved? • Perform an initial broad-scale analysis followed by a local analysis of the residuals • Explore 2DVAR in next two years. • More cross validation is required by withholding data • HadISST1 updated in near real time shortly put onto a password protected ftp site
AGCM vs CGCM • Power spectra of zonally averaged ocean surface air temperatures are not systematically different in Hadley CGCM and AGCM (forced with CGCM SSTs). Appears to be true for decadal time scales. • NCAR are running same experimental design to test if this result is model dependent • This may be incomplete because some modes which occur in CGCMs are missing from AGCMs, e.g. AGCMs may miss the covariance of monsoon rainfall and Pacific SSTs
Roles of AGCMs and CGCMs • CGCMs • investigate the nature of low frequency variability, e.g. THC (could bring in other participants). This can illuminate methods of analysing special events in AGCMs. It provides a qualitative difference from AMIP. • some covariability may be not captured by AGCMs • could nudge SSTs towards observed to diagnose limitations in AGCM (later). • AGCMs • can isolate variability completed forced by SST • able to investigate/understand specific historical events using many member ensembles
Additional data sets • Daily snow depth for 1977, 1979 onwards will be available once quality controlled. Should be ready for second stage of C20C.
Links to other groups • Coupled model working group • must not repeat their work • need to interface with them (Mitchell) • International CLIVAR group for seasonal to interannual prediction (WGSIP) (Zebiak). • - This is the more important link • get them to announce C20C experiments, leading way for others to participate. • should run past them a draft of CLIVAR exchanges report to bring them on board and include the link in the report explicitly • they may have suggestions on special topics.